Corruption risks, intensity of competition and estimated direct social loss in public procurement of Zagreb

Transcription

1 Corruption risks, intensity of competition and estimated direct social loss in public procurement of Zagreb Final report Budapest, September

2 The Corruption Research Center Budapest (CRCB) was created in November 2013 in response to the growing need for independent research on corruption and quality of government. Hence, the Center was established as a non-partisan research institute independent of governments, political parties or special interest groups. The aims of the Center are to systematically explore the causes, characteristics, and consequences of low quality of government, corruption, and regulatory failure using an inter-disciplinary approach. The Center also aims to help citizens to hold governments accountable through the use of robust evidence. Supporting partner: 3gteam: Corruption risks, intensity of competition and estimated direct social loss in public procurement of Zagreb The research was commissioned for the Centre for Peace Studies, Zagreb. The research was part of project under grant awarded by Central financing and contracting Agency (CFCA) of Republic of Croatia, reference: EuropeAid/135874/ACT/ID/HR, Building Local Partnerships for Open Governance and Fight against Corruption in Responsible Management of Natural Resources, project title : PINS II, implemented by Centre for Peace Studies. Authors: Tóth, István János - Hajdu, Miklós - Purczeld, Eszter Staff: Jancsó, Adrienne Kollár, Sándor Markson, Samuel Molnár, Balázs Orbán, Júlia Purczeld, Eszer Sáfár, Tamás Tóth, István János Tóth, Roland Ungár, Klára Vámos, Erika contributor research assistant physicist sociologist contributor research assistant research assistant economist & sociologist research assistant economist contributor Experts: Andor, Katalin, Goldstein, Katalin Gyenese, Jenő József, Magda Kelemen, Zoltán Székely, Attila economist language consultant software engineer lawyer lawyer procurement specialist Head of research: Tóth, István János The date of publication: September Corruption Research Center Budapest info@crcb.eu internet: 2

3 Executive Summary The research carried out by CRCB for the CMS is based on the analysis of 5,922 contracts of 4,483 public procurement issued by Grad Zagreb and Zagreb Holding between 2011 and During the data extraction, we could identify 1,197 winner companies on these tenders which are also the subject of the investigation. The following questions will be examined in the study: (i) What are the tendencies regarding the strength of competition and the corruption risks during the analysed time period? (ii) Which companies or groups of companies were the most successful on the tenders? Does the public procurement won by these companies differ from the rest by the corruption risks and the strength of competition? (iii) How are public procurement affected by the election years? How do the corruption risk and the intensity of competition in public procurement change in pre-election and in election years? (iv) To what extent can we estimate the direct social losses due to corruption and low intensity of competition in the period of ? How does the direct social losses differ between the two analysed issuer (i.e. Grad Zagreb and Zagreb Holding?) In the analysis we are using the Big Data approach to download all public procurement data of the aforementioned issuers from the official portal of the Croatian Public Procurement Authority. After the data extraction, we put the main information of all tenders to a structured database. We expand this database with company level data (ownership, personal ties and balance sheet data) from the Bisnode database. Then, after data cleaning, using statistical methods we analyse the corruption risks and intensity of competition from the aspect of the public procurement tenders and the winner companies as well. The total sum of net contract values suddenly increased between 2011 and 2013 (from 678 to 6,362 million HRK). Since a sudden decrease between 2013 and 2014 (from 6,362 to 3,480 million HRK) no clear tendencies or breakpoints could emerge until There are no clear breakpoints and permanencies in the yearly lists of the companies realising the biggest incomes from the public procurement of the City of Zagreb and Zagreb Holding. Tenders with single bidder generally involved companies that have less success in other procurement between 2011 and The corruption risk of public procurement increased significantly during the period. The share of tenders without competition increased from 25% to 34% between 2011 and Regarding the share of tenders without competition and comparing Zagreb's data with data of other European capitals, it can be seen that in Zagreb the 3

4 share of non-competitive tenders is much higher than in Ljubljana, Prague, Paris or especially in Vienna or Amsterdam. In , the tenders launched by Zagreb Holding had higher corruption risk than the tenders of the City of Zagreb. Since 2013, the City of Zagreb is the one with public procurement with higher corruption risks. During the period the intensity of competition decreased considerably at public procurement of the City of Zagreb. The results point out that the prices of public procurement are remarkably more distorted when there is no competition compared to those successful tenders with competition. The strength of price distortion increases significantly with the increase of corruption risk. The net contract prices of public procurement launched by Zagreb Holding are remarkably more distorted then ones of the City of Zagreb. There is a significant difference in price distortion among the contract prices in each year. While in 2013 (the year of the previous local elections) and in 2015 the first digits of net contract prices are very far from the expected (theoretical) distribution, in 2011, 2012, 2014 and in 2016 they fit well. The results pointed out that 27% of the total amount of money spent by public procurement was spent without competition. This high level shows that in Zagreb between 2011 and 2016 the competition practically did not exist in more than the quarter of the cases where public money was spent on public procurement. The amount spent without competition is considerable, approximatively 5 billion HRK in the whole period. The highest amount was spent without competition in 2013 by the Zagreb Holding (1.9 billion HRK). The median level of rate of direct social loss related to net contract value (DSLR) has moved between 31-35% during the whole period. The tenders launched by Zagreb Holding had higher median rate of direct social loss in 2011 (36-39%) and lower ones between 2013 and 2016 (28-33%). According to the method used we estimate that the total direct social loss in the whole period reached 1.47 billion HRK in public procurement of Zagreb Holding and 1.23 billion in tenders of City of Zagreb respectively. We estimate the highest amount of DSL, 813 million HRK in 2013 at tenders launched by Zagreb Holding. At tenders launched by the City of Zagreb the direct social loss increased from 90 million HRK (in 2011) to 419 million HRK in 2016, one year before the local elections. Concerning the economic branches the most suspicious tenders were in the IT sector. Here was the highest rate of tenders without competition; the indicators of price distortion showed high level of distortion: the highest level of rounded price and the highest level of declination of the distribution 4

5 of first digits from the theoretical distribution, and finally highest level of rate of direct social loss. Comparison of public procurement data of some European capitals on corruption risks and intensity of competition points out considerable differences. The results point out that Zagreb has the worst figures amongst European capitals concerning corruption risks and intensity of competition of public procurement. The analysis of public tenders launched by the Zagreb City and Zagreb Holdings in the period of points out that these public tenders were characterized by high corruption risks and low intensity of competition. As a result the social loss is significant. The analysis also points out that a group of Croatian companies are likely to incorporate the above mentioned characteristics of procurement procedures into their expectations, and tenders with low intensity of competition and high corruption risk play an important role in their business strategy. Our results also underline the need for a regular empirical analysis of the intensity of competition and corruption risks of Croatian public procurement - this could be the first step towards an increase of social welfare. 5

6 Contents Executive Summary... 3 Contents... 6 Introduction General tendencies between 2011 and Analysis of the winner companies Corruption risks and intensity of competition Price distortion Rounded data in contract prices Analysis of the first digits Estimation of direct social loss Money spending without competition Analysis of relative price drop to estimated value Corruption risks: an analysis at the winner company level References Abbrevations Appendix 1: Problems and errors of the official data publication Appendix 2: Distribution of main variables Appendix 3: The list the most important winners Appendix 4: The list of single bidding companies Appendix 5: Ownership networks Appendix 6: Personal networks

7 Introduction The research carried out by CRCB for the CMS is based on the analysis of 5,922 contracts of 4,483 public procurement issued by Grad Zagreb and Zagreb Holding between 2011 and During the data extraction, we could identify 1,197 winner companies on these tenders which are also the subject of the investigation. The following questions will be examined in the study: 1. What are the tendencies regarding the strength of competition and the corruption risks during the analysed time period? 2. Which companies or groups of companies were the most successful on the tenders? Does the public procurement won by these companies differ from the rest by the corruption risks and the strength of competition? 3. How are public procurement affected by the election years? How do the corruption risk and intensity of competition in public procurement change in pre-election and in election years? 4. To what extent can we estimate the direct social losses due to corruption and low intensity of competition in the period of ? How does the direct social losses differ between the two analysed issuer (i.e. Grad Zagreb and Zagreb Holding?) We are using the Big Data approach to download all public procurement data of the aforementioned issuers from the official portal of the Croatian Public Procurement Authority ( After the data extraction, we put the main information of all tenders to a structured database. We expand this database with company level data (ownership, personal ties and balance sheet data) from the Bisnode database. Then, after data cleaning 1 using statistical methods we analyse the corruption risks and intensity of competition from the aspect of the public procurement tenders and the winner companies as well. 1 For details, see Appendix 1 on issues concerning the data published by the Croatian Public Procurement Authority. 7

8 Pcs 1. General tendencies between 2011 and 2016 Regarding the number of contracts, a major decrease can be observed between 2013 and The total number of contracts was moving between 800 and 1216 per year between 2011 and 2016 (see Fig. 1.1.). Figure 1.1.: Yearly number of contracts between 2011 and 2016, N = 5, Source: CRCB own calculation based on data of EPRCRC 8

9 Million HRK The total sum of net contract values suddenly increased between 2011 and 2013 (from 678 to 6,362 million HRK). Since a sudden decrease between 2013 and 2014 (from 6,362 to 3,480 million HRK) no clear tendencies or breakpoints could emerge until 2016 (see Fig. 1.2.). We have to note that there were local elections in 2013 and also there will be such elections in 2017 a rise in the yearly sum of the net contract values can be expected for the years of the elections. 2 Figure 1.2.: Aggregated net contract values per year between 2011 and 2016, million HRK, N = 5, Source: CRCB own calculation based on data of EPRCRC 2 This assumption matches with the scientific results concerning the effects of electoral cycles and public expenses. Several papers investigate this topic on the macro level for instance see: Belo, et al. 2013; Bove, et al The present study points out such effects on the micro-level (i.e. on the level of public procurements) what can be regarded as a novelty on this field. 9

10 Most of the tenders were issued by Grad Zagreb in nearly every analysed year (the only exception was 2013); the ratio of contracts linked to Zagreb Holding was extremely low (1%) in 2016 (see Fig. 1.3.), because from 2016 City of Zagreb acts as a central body for public procurement of Zagreb Holding. Figure 1.3.: Distribution of contracts by issuer between 2011 and 2016, %, N = 5, % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 2% 28% 20% 41% 44% 57% 98% 72% 80% 59% 56% 44% Grad Zagreb Zagreb Holding Source: CRCB own calculation based on data of EPRCRC 10

11 Million HRK Even though the procurement between the two issuers was nearly equally distributed in 2013, the sum of the values of the contracts issued by Zagreb Holding significantly exceeds the sum that stands for the contracts of Grad Zagreb during this year that may be affected by the influences of the local elections in 2013 (see Fig. 1.4.). Figure 1.4.: Aggregated net contract values per year by issuer between 2011 and 2016, million HRK, N = 5, Grad Zagreb Zagreb Holding Source: CRCB own calculation based on data of EPRCRC In 2011, all of the contract values were expressed without the inclusion of VAT, but after a switch in 2012, all of the contract values include the VAT since 2013 (see Fig. 1.5.). Figure 1.5.: Distribution of contracts by VAT in their values between 2011 and 2016, N = 5, % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 59% 100% 100% 100% 100% 100% 41% 0% Without VAT With VAT Source: CRCB own calculation based on data of EPRCRC 11

12 The share of contracts deriving from open procurement procedures was constantly high during the observed period (87%-94%) (see Fig. 1.6.). Figure 1.6: Share of contracts deriving from open procurement procedures between 2011 and 2016, N = 5, % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 87% 89% 91% 88% 90% 94% 14% 11% 9% 12% 10% 6% Not open Open Source: CRCB own calculation based on data of EPRCRC Notes: 1) The following types of procurement were considered as open: Otvoreni and Otvoreni postupak. 2) The following types of procurement were considered as not open: Pregovarački bez prethodne objave, Pregovarački postupak bez prethodne objave, Pregovarački postupak s prethodnom objavom, Sklapanje ugovora bez prethodne objave poziva na nadmetanje (u slučajevima navedenim u Odjeljku 2 Priloga D1), Sklapanje ugovora o javnim uslugama iz Dodatka II.B.. The distribution of contracts between the different sectors shows considerable variability during the analysed time period; the most dominant sectors are the construction (23%-49%), the industry (14%-50%) and the area of other services (13%-34%) (see Fig. 1.7.). 12

13 Figure 1.7.: Distribution of contracts by sector between 2011 and 2016, N = 5, % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 15% 17% 13% 25% 23% 34% 9% 15% 11% 10% 11% 1% 24% 26% 27% 23% 35% 49% 50% 39% 37% 38% 26% 14% Industry IT Engeneering, RD, financial services Construction Real estate & services Other services Source: CRCB own calculation based on data of EPRCRC 13

14 2. Analysis of the winner companies There are no clear breakpoints and permanencies in the yearly lists of the companies realising the biggest incomes from the public procurement of Grad Zagreb and Zagreb Holding 3. The TOP15 lists for the most significant winner companies 4 are barely overlapping with each other. Table 2.1. shows the ratios how the TOP15 lists 5 correspond between the analysed years major overlap can be seen only between 2013 and 2015 (33%). Table 2.1: Ratio of correspondence between the lists of the TOP15 winner companies between the analysed years (2011 and 2016) % 27% 20% 20% 7% 13% % 100% 20% 13% 7% 20% % 20% 100% 27% 33% 20% % 13% 27% 100% 0% 20% % 7% 33% 0% 100% 7% % 20% 20% 20% 7% 100% Source: CRCB own calculation based on data of EPRCRC There are no companies that appeared continuously on the TOP15 lists of the analysed years. However, there are several companies that enter the TOP15 list in different, non-consecutive years: TEHNIKA d.d. (2011 and 2016) IKOM d.o.o. (2011 and 2014) USLUGA d.o.o. (2012 and 2016) PUGAR d.o.o. (2013 and 2015) VODOTEHNIKA d.d. (2012, 2014 and 2016) KONČAR - ELEKTRIČNA VOZILA d.d. (2013 and 2015) GUT d.o.o. (2013 and 2015) 3 Only the pubic procurements with one winner are taken into account in this chapter (N=3,923) as there is no information available in the data we had extracted about how the contract value was divided between multiple winners. 4 Winner companies with the highest aggregated net contract values were considered as the most successful ones in every analysed year. 5 See the lists themselves in Table

15 TIGRA d.o.o. (2013 and 2015) INA INDUSTRIJA NAFTE d.d. (2013 and 2016) HEP-OPSKRBA d.o.o. (2014 and 2016) In addition, some companies appear on the TOP15 lists in only two consecutive years: MEŠIĆ COM d.o.o. (2011 and 2012) HM-PATRIA d.o.o. (2011 and 2012) TEMEX d.o.o. (2012 and 2013) Furthermore, a few companies appear on the lists in both consecutive and non-consecutive years: PRIVREDNA BANKA ZAGREB D.D. (2011, 2013 and 2014) PETROL d.o.o. (2013, 2014 and 2016) Finally, ZAGREBAČKA BANKA d.d. is the only company that appear on the lists of several consecutive years as it was among the TOP15 winner companies between 2011 and Also, it is worth to highlight that GEORAD d.o.o. appears on the top lists between 2011 and 2013 and also between 2015 and

16 Table 2.2.: The TOP15 winner companies based on the aggregated net contract values per year between 2011 and 2016 Rank TEHNIKA d.d. ( HRK) ZAGREBAČKA BANKA d.d. ( HRK) VODOPRIVREDA ZAGREB d.d. ( HRK) PRIVREDNA BANKA ZAGREB D.D. ( HRK) MEŠIĆ COM d.o.o. ( HRK) IKOM d.o.o. ( HRK) DEKRA ZA PRIVREMENO ZAPOŠLJAVANJE d.o.o. ( HRK) GIP PIONIR d.o.o. ( HRK) HM-PATRIA d.o.o. ( HRK) V GRUPA d.o.o. ( HRK) GOLUBOVEČKI KAMENOLOMI d.o.o. ( HRK) SPEKTAR GRADNJA d.o.o. ( HRK) TEH-GRADNJA d.o.o. ( HRK) GEORAD d.o.o. ( HRK) UPS AGENCIJA ZA PRIVREMENO ZAPOŠLJAVANJE d.o.o. ( HRK) APIS d.o.o. ( HRK) USLUGA d.o.o. ( HRK) VODOTEHNIKA d.d. ( HRK) GEORAD d.o.o. ( HRK) ZAGREBAČKA BANKA d.d. ( HRK) MEŠIĆ COM d.o.o. ( HRK) AMB GRADNJA d.o.o. ( HRK) HM-PATRIA d.o.o. ( HRK) INSTAL PROM d.o.o. ( HRK) NERING d.o.o. ( HRK) KINDER GRADNJA ( HRK) PLANGRAD d.o.o. ( HRK) TEMEX d.o.o. ( HRK) SKEN-MONT d.o.o. ( HRK) HIDROCOMMERCE d.o.o. ( HRK) UniCredit Leasing Croatia d.o.o. ( HRK) ZAGREBAČKA BANKA d.d. ( HRK) ERSTE & STEIERMARKISCHE S-LEASING d.o.o. ( HRK) HYPO ALPE ADRIA LEASING d.o.o. ( HRK) PUGAR d.o.o. ( HRK) KONČAR - ELEKTRIČNA VOZILA d.d. ( HRK) GEORAD d.o.o. ( HRK) HYPO-LEASING KROATIEN d.o.o. ( HRK) PETROL d.o.o. ( HRK) GUT d.o.o. ( HRK) TIGRA d.o.o. ( HRK) PRIVREDNA BANKA ZAGREB D.D. ( HRK) TEMEX d.o.o. ( HRK) INA INDUSTRIJA NAFTE d.d. ( HRK) ĆIBO-PROMET d.o.o. ( HRK) Source: CRCB own calculation based on data of EPRCRC Note: the aggregated net contract values are in parentheses LUKOIL CROATIA d.o.o. ( HRK) PETROL d.o.o. ( HRK) CRODUX DERIVATI DVA d.o.o. ( HRK) HEP-OPSKRBA d.o.o. ( HRK) PRIVREDNA BANKA ZAGREB D.D. ( HRK) ERSTE & STEIERMÄRKISCHE BANK d.d. ( HRK) VODOTEHNIKA d.d. ( HRK) GRADSKA PLINARA ZAGREB - OPSKRBA d.o.o. ( HRK) ZAGREBAČKA BANKA d.d. ( HRK) IKOM d.o.o. ( HRK) KING ICT d.o.o. ( HRK) ERSTE & STEIERMARKISCHE S-LEASING d.o.o. ( HRK) BIROMAX d.o.o. ( HRK) RAIFFEISEN LEASING d.o.o. ( HRK) DIJANEŽEVIĆ AUTOPRIJEVOZ I GRADNJA d.o.o. ( HRK) KONČAR - ELEKTRIČNA VOZILA d.d. ( HRK) PUGAR d.o.o. ( HRK) P.G.P. d.o.o. ( HRK) TEGRA d.o.o. ( HRK) GEORAD d.o.o. ( HRK) GUT d.o.o. ( HRK) ŠUŠKOVIĆ-GRAĐENJE d.o.o. ( HRK) GTM d.o.o. ( HRK) PRIGORAC GRAĐENJE d.o.o. ( HRK) TIGRA d.o.o. ( HRK) EKO-MIKS d.o.o. ( HRK) HVAR d.o.o. ( HRK) M SOLDO d.o.o. ( HRK) NISKOGRADNJA DONJI JALŠEVAC d.o.o. ( HRK) HP-HRVATSKA POŠTA d.d. ( HRK) INA INDUSTRIJA NAFTE d.d. ( HRK) PETROL d.o.o. ( HRK) HEP-OPSKRBA d.o.o. ( HRK) TEHNIKA d.d. ( HRK) ELECTUS DGS d.o.o. ( HRK) USLUGA d.o.o. ( HRK) PROJEKTGRADNJA d.o.o. ( HRK) VIADUKT d.d. ( HRK) DUKAT d.d. ( HRK) PI VINDIJA d.d. ( HRK) GEORAD d.o.o. ( HRK) METRONET TELEKOMUNIKACIJE d.d. ( HRK) MONTER STROJARSKE MONTAŽE d.d. ( HRK) VODOTEHNIKA d.d. ( HRK) ŠKOLSKA KNJIGA d.d. ( HRK) 16

17 There are no systematic differences regarding the ratio of tenders won by the top winner companies between the groups of procurement with single and several bidders (see Fig. 2.1.) during the analysed time period. In 2013 and 2014, the top winners tended to win more of the tenders with single bidder (in 2013 and 2014). Conversely, in 2011, 2012, 2015 and 2016, the presence of the top winners was more prevalent among the tenders with several bidders. Figure 2.1.: Proportion of tenders won by the greatest winners in the given years by single bidder (SB), between 2011 and 2016, N=5, % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 13% 12% 8% 6% 7% 3% 6% 8% 7% 5% 7% 2% SB=0 SB=1 Source: CRCB own calculation based on data of EPRCRC SB = 0, more than one bidder = 1, only one bidder 17

18 Considering the ratio between the sum of the net values of the public procurement won by the companies belonging to the TOP15 lists and total aggregated values in analysed years, we can see a peak (55%) in 2014, in the year after the last local government elections; this finding suggests that in that year the money transferred via public procurement became more concentrated to the most significant winners. Figure 2.2.: Ratio of sum of net contract values of tenders won by companies belonging the TOP15 greatest winners and the total aggregated net contract values per year between 2011 and 2016, N=5, % 90% 80% 70% 60% 55% 50% 40% 41% 40% 43% 40% 43% 30% 20% 10% 0% Source: CRCB own calculation based on data of EPRCRC Note: the groups of the greatest winners are based on the net values of the public procurement aggregated to the level of winners in every year see Table 2.2 for the lists. In addition, we evaluated the performance of all of the winner companies on the public procurement. We calculated two scores for all of the winners in each year. The first one is based on the ratio between the number of tenders they had won and the total number of procurements in the given years. As for the second score, we calculate the ratio between the amount of money the given companies won on the tenders and total aggregated net contract values of the public procurement in every analysed year. Higher values of the scores indicate that a certain company won more tenders (or more money on tenders), so the higher scores mean high level of homogeneity and high level of concentration and the lower scores mean more heterogeneity and low level of concentration. For the sake of better interpretability, instead of the aforementioned ratios we publish the Z- 18

19 Points (mean) transformed 6 variants of these indicators 7. The indicator based on the number of tenders suggests that the winners of the public procurement had become less fragmented or in other terms less diverse between 2011 and 2013 (see Fig. 2.3.). This tendency reversed after However, the scores concerning the value of the tenders show a decreasing tendency during the analysed time period. This result points out that the distribution of the money on the tenders launched by the two issuers (Zagreb Holding and City of Zagreb) became less concentrated. Figure 2.3.: Average scores measuring winner companies performance on public procurement between 2011 and 2016, N=5, Score concerning the number of tenders Score concerning the value of tenders Source: CRCB own calculation based on data of EPRCRC Note: the higher values indicate the better performance on public tenders. Furthermore, in the case of tenders with single bidder these average performance scores tend to be lower between 2011 and 2015 (see Fig. 2.4.). Therefore, it can be concluded, that on tenders without competition the companies with lower performance on public procurement tended to win during this time period; in other words, the tenders with single bidder generally involved winners that had less success on other procurement. Although, regarding 2016, the opposite conclusion can be drawn, as the mean public procurement performance scores of the winners were higher 6 Z-transformation makes indicators more understandable and comparable. For details, see: 7 As these indicators characterize the winner companies, the standardization was done on the aggregation level of companies. Therefore, the distributions of the scores is not standard normal on the aggregation level of public procurements. 19

20 Points (mean) in the cases of the tenders without competition. Figure 2.4.: Average scores measuring winner companies performance on public procurement by number of bidders, between 2011 and 2016, N=5, SB=0SB=1SB=0SB=1SB=0SB=1SB=0SB=1SB=0SB=1SB=0SB= Score concerning the number of tenders Score concerning the value of tenders Source: CRCB own calculation based on data of EPRCRC Notes: 1) the higher values indicate the better performance on public tenders. 2) SB=0 indicates more than one bidder, SB=1 indicates single bidder. 20

21 3. Corruption risks and intensity of competition In this section, first we focus on the measurement and the analysis of corruption risks of public procurement tenders and then we deal with how the intensity of competition changed over the analysed period. The share of single bidder contracts is one other important indicator of corruption risk. The study of corruption risks is the study of the conditions of corruption. If somebody wants to be corrupt, then he/she sets up conditions to generate corruption. The corruption risk means that these conditions for corruption exist in the examined public procurement. The analysis of corrupt and collusive behaviour with hard data is an important new approach in the empirical research dealing with public procurement. In this report, we measure the corruption risk using an indicator which indicates the lack of competition during the tenders: there was only one bidder in the tender. Measuring the prevalence of single bidder contract we constructed an indicator Single Bidder (SB) using the following rule: SB = 1 if the tender was conducted with only one bidder SB = 0 if there were more than one bidder. In the tenders launched by the City of Zagreb and Zagreb Holding the share of tender with single bidder, i.e. tenders without competition, raised significantly, 9 percentage points between 2011 and 2016 (see Fig. 3.1 and 3.2.). This is a solid mark of rising tendency of corruption risk over the period. There is significant difference amongst European countries in this regard. Budapest performs better than Zagreb based on the national public procurement data: the former has less tenders without competition than the latter (see Fig. 3.3.). Regarding the data based on the European TED database which contains only tenders with large contract values we have to point out that the Zagreb s figures are better than the figure of Warsaw and much weaker than the figures of Ljubljana, Prague, Budapest or Rome and especially Paris, Vienna or Amsterdam (see Fig and Fig. 3.5.). In the latter three capitals the share of tenders without competition varied between 2 and 15 percent in the period of

22 Figure 3.1.: Share of tenders without competition (SB) by quarter, , %, N = 5, Source: CRCB own calculation based on data of EPRCRC Figure 3.2.: Share of tenders without competition (SB) by year, , %, N = 5, Source: CRCB own calculation based on data of EPRCRC 22

23 Figure 3.3.: Share of tenders without competition (SB) in Zagreb and in Budapest, , %, N = 5,922 (for Zagreb) and N = 2,849 (for Budapest) Budapest Zagreb Source: CRCB own calculation based on data of EPRCRC and MaKAB Figure 3.4.: Share of tenders without competition (SB) in Europe and in several European capitals, %, between , N = 3,407,027 Europe Warsaw Zagreb Prague Budapest Ljubljana Rome Berlin Vienna Paris Amsterdam Source: CRCB own calculation based on TED database 23

24 Figure 3.5.: Share of tenders without competition (SB) by year in several European capitals, , %, N = 3,407, Rome Paris Berlin Vienna Amsterdam Zagreb Budapest Prague Ljubljana Warsaw Source: CRCB own calculation based on TED database At the beginning of the period, in , there was a large difference between the public tenders of the City of Zagreb and Zagreb Holding regarding corruption risks, as the latter performed much worse. However, as single bidding became rarer in the public procurement of Zagreb Holding, and in the meantime, the City of Zagreb can be characterized by opposite tendencies, since 2013 the City of Zagreb is the one with public procurement with higher corruption risks. By the end of the period, the difference became large again between the two issuers (see Fig. 3.6.). 24

25 Figure 3.6.: Share of tenders without competition (SB) by year and issuer, , %, N = 5, Grad Zagreb Zagreb Holding Source: CRCB own calculation based on data of EPRCRC Amongst economic branches the share of tenders without competition is the highest in the IT sector and it is the lowest in the construction (see Figure 3.7.). 25

26 Figure 3.7.: Share of tenders without competition (SB) by year and sector, , %, N = 5, industry construction IT other services Source: CRCB own calculation based on data of EPRCRC Deriving information form the number of bidders b, we constructed two indicators which measure the intensity of competition (Indicator of Competitive Intensity) 8. Two indicators were defined in the following ways (see Table 3.1.). 8 See: CRCB, 2016 and Tóth & Hajdu

27 Table 3.1.: The definition of Indicator of Competitive Intensity (ICI and ICI2) condition function ICI if b = 2 ICI = lg2 if b = 3 or b = 4 ICI = lg[(3+4)/2] if b = 5 or b = 6 ICI = lg[(5+6)/2] if b = 7 or b = 8 ICI = lg[(7+8)/2] if b > 8 ICI = 1 if b = 1 ICI = 99, missing value ICI2 if b = 2 ICI = lg2 if b = 3 or b = 4 ICI = lg [(3+4)/2] if b = 5 or b = 6 ICI = lg [(5+6)/2] if b > 6 ICI = 1 if b = 1 ICI = 99, missing value Note: b: number of bidders The table 3.2. shows the distributions of tenders launched by the City of Zagreb and Zagreb Holding by value of indicators of competitive intensity (ICI and ICI2). 27

28 Table 3.2.: The distribution of tenders by the Indicator of Competitive Intensity (ICI and ICI2) and issuers, , N = 4,238 Grad Zagreb Issuers Zagreb Holding Total ici,30 Count ici2,30 % within issuer 24,7% 22,9% 24,1%,54 Count % within issuer 32,9% 36,1% 34,0%,74 Count % within issuer 23,8% 22,5% 23,4%,88 Count % within issuer 11,6% 10,8% 11,3% 1,00 Count ,54,74 1,00 % within issuer 7,0% 7,7% 7,2% Count % within issuer 24,7% 22,9% 24,1% Count % within issuer 32,9% 36,1% 34,0% Count % within issuer 23,8% 22,5% 23,4% Count % within issuer 18,6% 18,5% 18,6% Total Count % within issuer 100,0% 100,0% 100,0% Source: CRCB own calculation based on data of EPRCRC During the period the intensity of competition decreased considerably from 0.65 to 0.55 (See Fig. 3.8.). 28

29 Figure 3.8.: The average value of Indicator of Competitive Intensity (ICI) by year, , N = 4, Source: CRCB own calculation based on data of EPRCRC At the beginning and at the end of the period (in and 2016), the intensity of competition was higher at the tender launched by City of Zagreb than Zagreb Holding. However, between 2013 and 2015, the level of the competition intensity became higher in the case of the public procurement of Zagreb Holding (see Fig. 3.9.). Figure 3.9.: The value of Indicator of Competitive Intensity (ICI) by year and issuer, , N = 2, Grad Zagreb Zagreb Holding Source: CRCB own calculation based on data of EPRCRC 29

30 There was significant difference in the intensity of competition amongst industrial sectors: at tenders in construction that was higher and at tenders in IT sectors was lower during the period except in 2016 (see Fig ). Figure 3.10.: The value of Indicator of Competitive Intensity (ICI) by quarter, , N = 4, industry construction IT other services Source: CRCB own calculation based on data of EPRCRC Note: There were only 24 tenders in the IT sector during the whole period When we take into consideration the composition effects of public tenders by contract value, sector, year of tender and we count Amsterdam s data as reference, we could properly compare the corruption risk and intensity of competition of public tenders launched by the EU capitals. In figure 3.11 we show the level of corruption risks against of intensity of competition (ICI) of several European capitals. The results point out that Zagreb has the worst figures amongst European capitals concerning corruption risks and intensity of competition of public procurement. 30

31 Figure 3.11.: Corruption Risks and Intensity of Competition in selected EU Capitals, , N = 3,407,027 0 Berlin -0.5 Vienna Intensity of competition (ICIO) Paris Rome Ljubljana Budapest -2 Prague Warsaw Zagreb Corruption Risks (SB) Source: CRCB own calculation based on TED database Note: the coefficients of logit and ordered logit are on the x and y axis. The estimation were controlled by sector, year, and the logarithm of contract value; the reference capital was Amsterdam 31

32 4. Price distortion In this section, we focus on the analysis of contract prices to detect price distortion or overpricing. We interpret the price distortion as a sign of corruption risk. We use two methods to detect this phenomenon: we analyse the rounded data in contract prices (i), the observed distribution of first digits of contract price against to Benford s distribution (ii). Rounded data in contract prices Rounded contract prices can be regarded as an indicator of existence of price distortion 9 and a sign of corruption risk. We constructed four indicators for this analysis: ROUND10, ROUND10 and ROUND100. We defined them in the following way: RND10 = 1, if the net contract price is divisible by 10 without remainder (the contract price is rounded), else 0 RND100 = 1, if the net contract price is divisible by 100 without remainder (the contract price is rounded at hundreds), else 0 RND1000 = 1, if the net contract price is divisible by 1000 without remainder (the contract price is rounded at thousands), else 0. PRM3_1 = 1, if the net contract price is divisible by 3 without remainder, else 0. 9 The analysis of rounded data is one of tool the tools of fraud analytics to detect irregarities in prices. See Miller, 2015, Nigrini, 2012 and Spann,

33 The share of rounded net contract prices dropped significantly between 2011 and 2013, and thereafter practically stagnated (see Fig and 4.2.). Figure 4.1.: Share of rounded net contract price by year, % , N = 5,922 40% 35% 30% 25% 20% 15% 10% 5% 0% rnd_10 rnd_100 rnd_1000 Source: CRCB own calculation based on data of EPRCRC The data show that in the public procurement launched by Zagreb Holding, the net contract prices were rounded more often than the prices of public procurement by the City of Zagreb in the most of the years. Therefore, we can assume a stronger price distortion of the former (see Fig. 4.2.). 33

34 Figure 4.2.: Share of rounded contract price by year and issuer, % , N = 5, Grad Zagreb, rnd_10 Grad Zagreb, rnd_100 Zagreb Holding, rnd_10 Zagreb Holding, rnd_100 Source: CRCB own calculation based on data of EPRCRC On the other hand, concerning the intensity of competition, it is important to note whether a local business (company from Zagreb) or a business from the countryside (outside Zagreb) was the winner. In the latter case, the existence of stronger competition in a wider market can be assumed. The net contract prices are less rounded where the winner was a business out of outside Zagreb, than in the cases of the companies price from Zagreb (See Fig. 4.3.). 34

35 Figure 4.3.: Share of rounded contract price by year and the seat of winning company, %, , N = 5, winner from Zagreb, rnd_10 winner from Zagreb, rnd_1000 winner from out of Zagreb, rnd_10 winner from out of Zagreb, rnd_1000 Source: CRCB own calculation based on data of EPRCRC Overall, it can be stated that the likelihood of rounded contract values and thus the likelihood of price distortion is greater in the tenders issued by Zagreb Holding and awarded by the Zagreb companies than other tenders. The ratio of rounded prices (with rounded by 100) in tenders launched by the Zagreb Holding was 10.8%, and by the City of Zagreb was 10.4%; and amongst the winners for companies from Zagreb was 11.5% and for the companies outside of Zagreb was only 8.4% (see Table 4.1.). The differences in the use of rounded prices are quite large by the level of intensity of competition: in tenders with high level of competition, the winners are rounded off at a lower rate than the tenders with high level of intensity of competition (8.6% against 11.5% for rounding 100). There are considerable differences in price distortions amongst economic branches: the use of rounded data in the IT sector is the most widespread, around half of the tenders net contract prices (45.3%) are rounded up to 10 kunas and nearly one-third are 29.3% to 100 kunas in the analysed period. 35

36 Table 4.1.: Share of rounded price by several group of tenders, %, , N = 5,922 round_10 round_100 round_1000 Issuer: Grad Zagreb 16.5 * Issuer: Zagreb Holding 18.9 * Winner company is from Zagreb 18.5 * 11.5 * 6.2 * Winner company is form outside of Zagreb 14.4 * 8.4 * 3.9 * Low intensity of competition (ICI=0.301) 17.8 * 11.4 * 7.2 * High intensity of competition (ICI=1) 13.5 * 8.6 * 3.2 * Industry 10.8 * 5.8 * 2.9 * Construction 6.5 * 1.7 * 0.7 * IT 45.3 * 29.3 * 15.3 * Other services 31.1 * 21.9 * 12.1 * Notes: *: the value of chi 2 is significant at p <0.05 level +: the value of chi 2 is significant at p <0.1 level Source: CRCB own calculation based on data of EPRCRC The estimation of odds of rounding prices and prices which can be divided into three shows that lack of competition or weak competition increases the chances that the net contract value contains some degree of rounding (See Table 4.2.). We can assume then that the highest level of rounding is a clear sign of the greater chance of price distortion. 36

37 Table 4.2.: Estimation of rounded price (ROUND_10 and ROUND_1000) and PRM3_1 by binary logistic estimation, ROUND_10 ROUND_1000 variables ICI -0,197-1,111* Winner: from Zagreb 0,048 0,074 Issuer: Zagreb Holding 0,206 0,188 Industry -1,185* -1,499* Construction -2,391* -3,351* IT 0,571 0,098 Other services ref. ref. Logarithm of net contract value (lnncvx) -0,122* -0, ,171* 1,990* ,015* 0,811* ,211-0,151* ,062 0, ,015-0, ref. ref. constant 0,190-1,569* N 4,185 4,185 Model Chi-square * Log likelihood Nagelkere R Square Notes: *: p < : p < 0.1 Source: CRCB own calculation based on data of EPRCRC 37

38 Analysis of the first digits Using the second method, we analyse the price distortion by the distribution of the first digit in the contract prices based on Benford s law 10. According to Benford's law (also known as the First-Digit Phenomenon) in a non-artificially generated set of numbers (in any numeral system) the first digits in each, local values are distributed neither arbitrarily nor uniformly; the distribution instead follows the distribution set by Benford s law 11. The distribution of first digits in the decimal system (1,..,9) according to Benford s law is in Table 4.3. Table 4.3.: The distribution of first digit according to the Benford s law in the decimal system First digit % The economist Hal Varian first suggested in 1972 that Benford s law could be used to detect possible fraud in socio-economic data, and that it the performance of forecasting models could be evaluated 12. Mark Nigrini pointed out 25 years later that Benford's Law is useful in forensic accounting and auditing as a tool to detect fraud and collusion 13. Ever since, Benford s Law has been common and it is a widely used method in several areas of social research for fraud detection 14. For the analysis of irregularities in public procurement, we can use the information on procurement prices because these are public (a); and as such these may carry information on the process of price formation (b). Our 10 In the description of the concept of this method for the detection of price distortion we are using partially our earlier work. See CRCB, A set of numbers is said to satisfy Benford's law if the leading digit d (in 10 digit system, d {1,..., 9}) occurs with probability: P (d) = log10 (d +1) - log10 (d) = log10 (1 + 1/d). See 12 See Varian, See Nigrini, 1996; Drake, Nigrini, 2000; Durtschi, et al., See Nigrini, 2012; Miller,2015; Kossovsky,

39 research questions related to the price formation are the following: whether the price formation differs significantly amongst different group of public procurement created by intensity of competition (i), the risks of corruption (ii) and the two issuers (City of Zagreb and Zagreb Holding). We examine these relationships with comparison of observed first digit s distribution to theoretical (Benford s) distribution of contact prices of tenders in several analysed groups of the Hungarian public procurement. The analysis of first digits indicates that the contract prices of all public procurement launched by the two issuers fit the theoretical distribution for the whole period ( ) (see Figure 4.4). Figure 4.4.: The expected and observed distribution of first digits in net contract value, %, , N = 5, obs exp Source: CRCB own calculation based on data of EPRCRC 39

40 There is significant difference in price distortion among the contract prices in each year. While in 2013 (the year of the previous local elections) and in 2015 the first digits of net contract prices are very far from the expected (theoretical) distribution, in 2011, 2012, 2014 and in 2016 they fit well (see Fig. 4.5.). Figure 4.5.: The weight of price distortion: the squared error (SE) of contract prices of PPZ from the theoretical (Benford s) distribution by year, , N = 5, Source: CRCB own calculation based on data of EPRCRC Note: the bars in red do not fit the expected (Benford s) distribution PPZ: Public Procurement of Zagreb Holding and City of Zagreb 40

41 The construction and the sector of other services have the smallest level of price distortion while in the IT sector the observed distribution of first digits has extremely high level of difference from the theoretical distribution (see Fig. 4.6.). Amongst the sectors, only the prices of IT sector do not fit the theoretical model 15. The high level of the former is certainly related to the high level of overpricing in this sector. Figure 4.6.: The weight of price distortion: the squared error (SE) of contract prices of PPZ from the theoretical (Benford s) distribution by sectors, , N = 5, industry construction IT other services Source: CRCB own calculation based on data of EPRCRC Note: the horizontal (x) axis: the first digits the vertical (y) axis: squared error of observed and expected percentiges: (obs exp)*(obs - exp) Our results also point out that the prices of public procurement are remarkably distorted when there is no competition compared to those successful tenders with competition. So, the results indicate that the strength of price distortion increases significantly with the increase of corruption risk (see Fig. 4.7.). 15 The MAD value is which is far over the threshold (0.012) suggested by Nigrini. 41

42 Figure 4.7.: The squared error between the expected (Benford's) and observed distribution of first digits by first digits and the indicator of corruption risk (SB), , N = 5, sb=0 sb=1 Source: CRCB own calculation based on data of EPRCRC Note: horizontal (x) axis: the first digits vertical (y) axis: squared error of observed and expected percentiges: (obs exp)*(obs - exp) And finally, our results also point out that the net contract prices of public procurement launched by Zagreb Holding are remarkably more distorted then onces of City of Zagreb (See Figure 4.8). 42

43 Figure 4.8.: The squared error between the expected (Benford's) and observed distribution of first digits by first digits and issuers, , N = 4, grad zagreb zagreb holding Source: CRCB own calculation based on data of EPRCRC Note: horizontal (x) axis: the first digits vertical (y) axis: squared error of observed and expected percentiges: (obs exp)*(obs - exp) 43

44 5. Estimation of direct social loss Two methods were used to estimate the direct social loss associated with public procurement corruption. First, we calculated the amount spent by the two issuers (City of Zagreb and Zagreb Holding) without competition, that is, with high corruption risks (i), then we used the net estimated value and net contract value and their difference as a tool, to estimate the strength of competition and the expected price drop relative to estimated value at tenders with high level of competition (ii). Money spending without competition For the first estimate, we simply calculated how much money was spent without competition. The larger the amount of money without competition is the greater the social loss is. The results point out that approximately 27% of the total amount of the money spent on public procurement was spent without competition during the whole period (see Fig. 5.1.). We have to consider that level quite high: in Zagreb between 2011 and 2016 the competition practically did not exist at more than the quarter of public money spent on public procurement. The highest value (43%) was in 2011 and the lowest one (21%) in Figure 5.1.: The share of money spent in PPZ without competition, %, , N = 5,922 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 43% 27% 33% 22% 21% 23% 27% Whole period Source: CRCB own calculation based on data of EPRCRC 44

45 According to these results, the amount spent without competition is considerable, approximatively 5 billion HRK in the whole period. The highest amount was spent without competition in 2013 by the Zagreb Holding (see Fig and 5.3.) Figure 5.2.: The sum of the value of PPZ without competition, in million HRK, , N = 1, Source: CRCB own calculation based on data of EPRCRC The values from 2013 are related to urban transport (bus and tram) purchases. The Zagreb Holding entrusted several banks with leasing services without competition (see tables A4.1. and A4.2. in Appendix 4). For these transactions the public tenders were organised strangely on the same day (i) and for the purchase of the same service (ii) and only one bidder participated in each (iii). 45

46 Figure 5.3.: The sum of the value of PPZ without competition by issuer, in million HRK, , N = 1, Grad Zagreb Zagreb Holding Source: CRCB own calculation based on data of EPRCRC Analysis of relative price drop to estimated value To estimate the direct social loss due to corruption, we calculated the magnitude of price drop of the contract price compared to the estimated value using the following formula: RPRD = (P P) P 100 (1) Where P * is the estimated net value and P is the net contract price of the tender. We consider this indicator as a new measure of intensity of competition: the greater value of RPRD (i.e. greater magnitude of price drop) is the higher level of intensity of competition can be traced in the public tenders. The low or zero value of RPRD means low level or lack of competition. The rate of price drop correlates strongly with the indicators of corruption risk and the intensity of competition. In tenders with low corruption risk (SB=0) and high level of intensity of competition (ICI) the net contract prices dropped significantly at a higher rate compared to the estimated price than where the corruption risks remained high and the intensity of competition was rather weak. 46

47 The concept of the estimation is based on the assumption that there is a chance that the corruption risk of any tender can stay low and the intensity of competition can reach a high level. Observing the rate of price drop in the tenders with low corruption risk and high level of competition we can mark out these high rates as reference points, as outcomes of the ideal or clean public procurement process. In this way we can estimate in case of every tender how much the estimated price should have dropped compared to this reference level. According to this concept we can estimate the rate of direct social loss in the given tender if we extract the observed rate of price drop (RPRDobserved) from the reference rate, which came from the ideal, non-corrupt cases, (RPRDreference): DSLR = RPRD reference RPRD observed So, for every i tenders, where we have data on RPRD, we calculate the rate of direct social loss (DSLRi) as follows: DSLR i = RPRD max RPRD i And the multiplication of the DSLRi by the net contract value (NCVi) of the i tender gives us the amount of social loss for every i tender. And finally, this way it could be easier to calculate the aggregate direct social loss for all n tenders: n DSL = i=1(dslr i NCV i ) Using this method we have to confront several limitations. First, the used method is incapable of detecting certain forms of corruption. Focusing on the relative price drop from the estimated price we could not detect the corruption cases which were related to so called white elephant projects and the social losses of these projects (i) 16. Second, the corruption indicators and proxies of intensity of competition which we have been using in the analysis certainly do not measure every form and type of corrupt activities (ii). Obviously, there are such forms of corrupt activities which are beyond our scope (i.e. collusion and bid rigging which are used very frequently in the construction sector). Using the concept presented above we calculated two estimations. In these 16 The concept of white elephant projects is well known in the corruption literature (Rose- Ackermann, 2006; Rose-Ackerman-Soreide, 2011). These are projects without any social benefit or those that are ruined shortly after their completion. 47

48 estimations we used different assumptions concerning the reference rate (RPRDmax) which happened in case of the ideal, non-corrupt public tenders. First we calculated the median value of RPRD for all tenders grouped by the level of corruption risk and then according the level of indicator of competitive intensity (ICI2), (See Table 5.1.). For the estimation we use the RPRDmax= value as the reference (benchmark) value. With this decision we assume that the tenders where there are competition (SB=0) are the normal solutions for public money spending, i.e. the tenders with low corruption risk are the normal against those where there was only one bidder (and thus the corruption risk was high). This assumption is much weaker than if we considered tenders with high intensity of competition to be normal", and we would have accepted the RPRD value at highest intensity of competition (79.1) as the reference value. Table 5.1.: The median value of RPRD at several group of tenders defined by corruption risks and intensity of competition, , N = 5,071 Group of tenders Median value of RPRD N tenders with competition (SB = 0) tenders without competition (SB = 1) low level of intensity of competition (ICI2=0.301) ICI2 = ICI2 = high level of intensity of competition (ICI2 = 1)

49 Figure 5.4.: The median value of the ratio of estimated direct social loss in net contract value by year, %, , N = 3, Source: CRCB own calculation based on data of EPRCRC Note: the reference value of relative price drop (RPRD) is Figure 5.5.: The median value of the ratio of estimated direct social loss (DSLR) in net contract value by level of competition, %, , N = 1, two bidders 3-4 bidders 5-6 bidders more than 6 bidders with competition without competition Source: CRCB own calculation based on data of EPRCRC Note: the reference value of relative price drop (RPRD) is The median level of DSLR has moved between 31-34% during the whole period (see Fig. 5.4.). The estimated direct social loss differs significantly amongst tenders defined by intensity of competition and corruption risk 49

50 (see Fig. 5.5.). We estimate the lower median of social loss at tenders with high level of competition, i.e. more than 6 bidders (22%) and the highest level at tenders where there was no competition (40%). The tenders launched by Zagreb Holding had higher median value in 2011 and 2012 (36-39%) and lower ones after 2013 (30-33% in comparison to Grad Zagreb (see Fig. 5.6.). Amongst the economic sectors, we can observe the highest level of DSLR in the IT sector (the median value was 40%), and the lowest one at construction (29%) (See Fig. 5.7.). 50

51 Figure 5.6.: The median value of the ratio of estimated direct social loss in net contract value by issuer, %, , N = 3, Grad Zagreb Zagreb Holding Source: CRCB own calculation based on data of EPRCRC Note: the reference value of relative price drop (RPRD) is

52 Figure 5.7.: The median value of the ratio of estimated direct social loss in net contract value by sector, %, , N = 3, industry construction IT other services Source: CRCB own calculation based on data of EPRCRC Note: the reference value of relative price drop (RPRD) is Finally, according to the method used we estimate that the total direct social loss in the whole period reached 1.47 billion HRK in public procurement of Zagreb Holding and 1.23 billion in tenders of City of Zagreb respectively. We estimate the highest amount of DSL, 813 million HRK in 2013 at tenders launched by Zagreb Holding (See Fig. 5.8.). Concerning the City of Zagreb the estimation shows a rising tendency of the weight of direct social loss from 90 million HRK in 2011 to 419 million HRK in 2016 one year before the local elections. 52

53 Figure 5.8.: The estimated direct social loss by issuer, million HRK, , N = 3, Grad Zagreb Zagreb Holding Source: CRCB own calculation based on data of EPRCRC Note: the reference value of relative price drop (RPRD) is

54 6. Corruption risks: an analysis at the winner company level The issuers (City of Zagreb and Zagreb Holding) have given the company tax number in 5,260 of the 5,922 public tenders analysed. In these cases we could look up the turnover data in the years from the database of Bisnode 17. (The data of 2016 were not available yet.). Thus, finally we get data on 1,040 of the 1,197 winner companies and in the analysis we used aggregated data of the whole analysed period. Our two main indicators were the followings: 1. TSLS_SUM: total net sales 2. EXSLS_SUM: total export sales And, we used four indicators from the public tender database created by CRCB: 3. NCVALUE_SUM: total net contract value 4. N_SUM: number of tenders won by the winner company 5. SB_SUM: number of tenders won by the winner company, there were only one bidder 6. DSL_SUM aggregate value of direct social loss in tenders won by the winner company In the next step we constructed the following four indicators at company level used the indicators 1-6: 7. The relative weight of total net contract value in total net sales: PPR = NCVALUE_SUM / TSLS_SUM * The share of export turnover in total net turnover: EXPR = EXSLS_SUM / TSLS_SUM * The weight of direct social loss in total contract value of all tenders won by the winner company: DSLR = DSL_SUM / NCVALUE_SUM * The share of tender won by the winner company as single bidder in total number of tenders won: AV_SB = SB_SUM / N_SUM * See 54

55 Our aims in this company-level analysis are to detect the links between the share of aggregated contract value of public tenders in the winner company s total net turnover and the weight of corruption risks at tenders won by this given company. If we detect positive correlation, either this points out that the appearance in the public procurement market is predetermined by the expectation that in this market the intensity of competition may be low and the corruption risks may be high, and finally, in such environment a company which uses these opportunities can win easily and it can quickly increase their net turnover. Or, such results can simply mean that the success of the winning companies, which can be measured here by the aggregated contract value related to the public tenders per total turnover ratio, is significantly supported by the special environment which the given company creates: weak competition and high risk of corruption. If there is no such relationship, this suggests that the fact of high corruption risk and low intensity of competition or expectations about them do not play a role in the strength of involvement of companies in the public procurement market. The other issue is the relationship between the ratio of direct social loss in aggregated contract value and the strength of involvement of winner company in the public procurement market. If in case of companies which connectivity to the public procurement market is relatively strong, and the ratio of direct social losses at tenders won by these companies is relatively high, then this relationship clearly implies the attractiveness of a special market segment (i.e. public procurement) which is characterised by high corruption risk and low level of intensity of competition. In addition, we can consider a clear sign of such relationship as the negative link between the strength of the presence on the public procurement market (PPR) and the share of export turnover (EXPR). The companies which export their products are exposed to stronger competition in the export markets than in the domestic market or the public procurement market. The negative link between the two indicators (PPR and EXPR) underlines that companies that are more present in export markets are less likely to turn to the public procurement market. In addition, companies with strongest involvement in the public procurement market export significantly less. This relationship may be merely a result of the sectoral effect (e.g. a construction company specializing in large projects has rarely high export rate) or explanations due to considerable expected difference regarding the intensity of competition between the two markets. 55

56 The distribution of PPR, EXPR, AV_SB, and DSLR is a power function-like, and therefore we calculated their values converted to ordinal variables. The results confirm the negative correlation between the export share and the weight of the presence in the public procurement market: in companies where the public procurement market plays a minor role the share of export in total turnover is higher than where public procurement plays highest role. The latter ones typically do not export or export only a small proportion (see Figure 6.1). Figure 6.1.: The share of export in total net turnover (EXPR) by weight of total contract value of public tenders in total net turnover (PPR), %, , N = expr: zero expr: no more than 10% expr: more than 10% ppr: no more than 10% ppr: more than 10% Source: CRCB own calculation based on data of EPRCRC Chi2 (2): , p < 0.01 The results also point to the existence of a positive link between corruption risks of public tenders and its importance in total sales of the company (see Figure 6.2). If the role of the public procurement market is greater within the company's sales, the company typically won public tenders characterized by high corruption risks. So either the anticipated or expected high corruption risk encourages the entry to the public procurement market for the future corrupt companies, or all companies are trying to reach the public procurement market, but the corrupt companies are far more successful and thus their involvement in this market segment is stronger compared to others. The positive relationship between direct social loss and the importance of public procurement market is also significant (Figure 6.3). There is no significant difference between winning companies regarding social losses 56

57 below 40%. But there is high difference where the direct social loss is higher than 40%. For companies with highest share of public procurement market (more than 10% of sales revenue), the share of tenders with a social loss more than 40% is much higher (24%), compared to those in which public procurement market has little role (11.5%). Consequently, the share of wasted money (social loss) is higher in public tenders won by companies strongly connected to the public procurement market. This phenomenon is clearly related to the weak competition and high corruption risks. Figure 6.2.: The share of tender won the winner company as single bidder (AV_SB) in total number of tender by the weight of total contract value of public tenders in total net turnover of the company (PPR), %, , N = avsb = 0 0 < avsb < 1 avsb = 1 ppr: no more than 10% ppr: more than 10% Source: CRCB own calculation based on data of EPRCRC Chi2 (2): 8.479, p < 0.05 Thus, companies are able to reduce competition and create favourable conditions for corrupt transactions for which public procurement is an important market, and the total contract value earned here gave more than 10% of their net sales between 2011 and The incentives for companies relying more on public procurement are stronger and their experience is greater in creating favourable environment for corruption than for companies that are only slightly involved in public procurement market (see Table 6.2.). The analysis of public tenders launched by the Zagreb City and Zagreb Holdings in the period of points out that these public tenders were characterized by high corruption risks and low intensity of competition. As a result the social loss is significant. The analysis of corruption risks and market orientation at winning company level also points out that a group of 57

58 Croatian companies is likely to incorporate the above mentioned characteristics of procurement procedures into their expectations, and tenders with low intensity of competition and high corruption risk play an important role in their business strategy. Our results also underline the need for a regular empirical analysis of the intensity of competition and corruption risks of public procurement - this could be the first step towards an increase of social welfare. Figure 6.3.: The rate of direct social loss (DSLR) by the weight of total contract value of public tender in total net turnover of the company (PPR), %, , N = dslr 10% 10%<dslr 20% 20%<dslr 30% 30%<dslr 40% 40%<dslr ppr: no more than 10% ppr: more than 10% Source: CRCB own calculation based on data of EPRCRC Chi2 (4): , p < 0.05 Table 6.1.: Ordered logit estimation of corruption risks (AVSBO), , N = 878 AVSBO variables PPRO=1 (PPR<10%) Ref. PPRO=2 (PPR >= 10%) N_SUM * Cut Cut N 878 Log likelihood LR Chi2 (2) Prob > chi Pseudo R Square Notes: *: p < : p < 0.1 Source: CRCB own calculation based on data of EPRCRC 58

59 References ACFE Report to the Nations on Occupational Fraud and Abuse, 2016 Global Fraud Study. Association of Certified Fraud Examiners, USA: Austin, Texas. Belo, F., Vito, D., Galab, V., Lic, J Government spending, political cycles, and the cross section of stock returns, Journal of Financial Economics, Volume 107, Issue 2, February 2013, Pages Bove, V., Efthyvoulou, G., Navas, A Political cycles in public expenditure: butter vs guns, Journal of Comparative Economics, Available online 1 April 2016, Chvalkovska, J., Fazekas, M., Skuhrovec, J., Tóth, I. J., King L. P Are EU funds a Corruption Risk? The Impact of EU Funds on Grand Corruption in Central and Eastern Europe. In: Pippidi-Mungiu, A. Controlling Corrution in Europe. The Anticorruption Report 2. Oplanden, Berlin & Toronto: Barbara Budrich Publishers. pp Czibik, Á., Fazekas, M., Tóth, B., Tóth I. J Toolkit for detecting collusive bidding in public procurement. With examples from Hungary. Working Paper Series: CRCB- WP/2014:02. CRCB, Budapest, Durtschi, C. - Hillison, W.- Pacini, C The Effective Use of Benford's Law to Assist in. Detecting Fraud in Accounting Data, Journal of Forensic Accounting, Vol V. pp , Drake, P. D. Nigrini, M. J Computer assisted analytical procedures using Benford s law, Journal of Accounting Education, Vol. 18. no. 2. pp ; Fazekas, M., King, L. P., Tóth, I. J Hidden Depths. The Case of Hungary. In: Pippidi- Mungiu, A. Controlling Corrution in Europe. The Anticorruption Report 1. Oplanden, Berlin & Toronto: Barbara Budrich Publishers. pp Fazekas, M., and Tóth, I. J From Corruption to State Capture. A New Analytical Framework with Empirical Applications from Hungary. Political Research Quarterly, June 2016, vol. 69. no. 2. pp , doi: / Fazekas, M., Tóth, I. J., King, L. P An Objective Corruption Risk Index Using Public Procurement Data. European Journal on Criminal Policy and Research, First Online: 25 April 2016 doi: /s z. Fazekas, M. - Tóth, I. J Corruption in EU Funds? Europe-wide evidence of the corruption effect of EU-funded public contracting. In: Bachler, J., Berkowitz, P., Hardy S., Muravska, T.: EU Cohesion Policy. Reassessing Performance and Direction, Routledge, London & New York., pp Kossovsky, A. E Benford s Law. Theory, the General Law of Relative Quantities, and Forensic Fraud Detection Applications. Hackensack, New Jersey, USA: World Scientific Miller, S. J. (ed.) Benford s Law: Theory and Applications. Princeton, New Jersey, USA: Princeton University Press 59

60 Nigrini, M. J A taxpayer compliance application of Benford s law. Journal of the American Taxation Association. Vol. 18. no 1. pp Nigrini, M. J. (ed.) Benford's Law. Applications for Forensic Accounting, Auditing, and Fraud Detection. Hoboken, New Jersey, USA: John Wiley & Sons Spann, Delena D Fraud Analytics: Strategies and Methods for Detection and Prevention, Hoboken. New Jersey, USA: John Wiley & Sons Szánto, Z., Tóth, I. J., Varga, S The social and institutional structure of corruption: some typical network configurations of corruption transactions in Hungary. In: Vedres, B., Scotti, M..eds. Network in Social Policy Problems. Cambridge, UK: Cambridge University Press. Rose-Ackerman, S. (ed.) International Handbook on the Economics of Corruption, Edward Elgar, Cheltenham, UK. Rose-Ackerman, S. Soreide, T International Handbook on the Economics of Corruption. Volume Two. Edward Elgar, Cheltenham, UK. Tóth, I. J., Hajdu, M Competitive Intensity and Corruption Risks in the Hungarian Public Procurement Paper presented at the University of Cambridge, Data for Policy Conference, Tóth, I. J., Hajdu, M Intensity of Competition, Corruption Risks and Price Distortion in the Hungarian Public Procurement The report prepared for the EU Comission. Varian, H. R.1972.: Benford s law, The American Statistician, 26. Vol. no.3. pp

61 Abbrevations CRCB DSL DSLR EPRCRC HRK ICI MAD MaKAB NCV RPRD SB TED PPZ Corruption Research Center Budapest Direct Social Loss Ratio of Direct Social Loss relative to Net Contract Value Electronic Public Procurement Classifieds of the Republic of Croatia Croatian Kuna Indicator of Competitive Intensity Mean Absolute Deviation Database of Hungarian Public Procurement Net Contract Value Relative price drop compared to the estimated value Public tender with single bidder Tenders Electronic Daily This is the online version of the 'Supplement to the Official Journal' of the EU, dedicated to European public procurement ( Public Procurement of Zagreb Holding and City of Zagreb 61

62 Appendix 1: Problems and errors of the official data publication The Croatian Public Procurement Authority published data about 4,653 public procurement issued by Grad Zagreb and Zagreb Holding in total between 2011 and As several contracts may belong to one procurement, our analysis was based on a contract level dataset, that contains 6,172 cases in total. However, during our analysis, we had to filter out 250 cases because 0 HRK was indicated as the contract value. We assume that these tenders were not valid, therefore we excluded them from the analysis. In addition, the net contract value was more than ten times higher than the estimated contract value in the case of 41 public procurement. As we suspect that this phenomenon occurred because of mistyping the decimal separator or some other kind of mistake during the publication of data, in these cases we taken into consideration the estimated contract value as the actual net contract value. Furthermore, we would like to indicate that there may be some additional misreported contract values based on the relation between the net contract values and the estimated values (see Table A1.1.). However, because there is no clear general justification for filtering out these tenders, we decided to keep them in the analysis. We suggest the one-by-one analysis of these cases when contract price was published faulty by the Croatian authorities (see them in Table A1.2.) as a further step of the research in order to reveal the reasons of the inconsistencies. 62

63 Table A1.1: Number of suspicious cases by year where the net contract value (NCV) exceeds the estimated value (EV) NCV is more than 1.5 times higher than EV NCV is more than 2 times higher than EV NCV is more than 3 times higher than EV NCV is more than 5 times higher than EV NCV is more than 10 times higher than EV Total Source: CRCB Note: the cases in bold are excluded from the analysis. 63

64 Table A1.2: List of cases where the net contract value (ncvalue) exceeds significantly the estimated value (c_value_est) id date_ issu er w_name c_value_es t ncvalue sb x2 x3 x PRONGRAD BIRO d.o.o Georad d.o.o. i Geodist d.o.o Siemens d.d PEEK PROMET d.o.o PEEK PROMET d.o.o PEEK PROMET d.o.o FANOS d.o.o. Za projektiranje i inženjering u prometu FANOS d.o.o SEMAFOR d.o.o USLUGA d.o.o BENUSSI d.o.o GRA-PO d.o.o Poljoopskrba tehno d.d Poljoopskrba tehno d.d Poljoopskrba tehno d.d Poljoopskrba tehno d.d AUTO-SAFIR d.o.o Finvest corp d.d Purić d.o.o R-PIM d.o.o Industrooprema d.o.o.; MIK-ELING d.o.o DOMEL d.o.o.; GRAĐPROM d.o.o GEORAD d.o.o P.G.P. d.o.o TEMEX d.o.o

65 TIGRA d.o.o PUGAR d.o.o Drager Safety d.o.o VUGRINEC d.o.o DETA PRUT d.o.o RESNIK-BETON d.o.o RO-TEHNOLOGIJA d.o.o Industrooprema d.o.o VODOSKOK d.d VODOSKOK d.d VODOSKOK d.d VODOSKOK d.d EOL GRUPA d.o.o GUT d.o.o O-K-TEH d.o.o VATRO PROMET d.o.o KEFO d.o.o SITOLOR d.o.o AUTO ENIGMA d.o.o AUTO-MAG d.o.o AUTO-MAG d.o.o AUTO-MAG d.o.o Auto - Mag d.o.o AUTO-MAG d.o.o ELEKTRO - KOMUNIKACIJE d.o.o GRA-PO d.o.o NERING d.o.o

66 SHIMADZU d.o.o SMIT - COMMERCE d.o.o SMIT - COMMERCE d.o.o SMIT-COMMERCE d.o.o SMIT-COMMERCE d.o.o URIHO Zagreb ZAGREBAČKO ELEKTROTEHNIČKO PODUZEĆE d.d ZAGREL RITTMEYER d.o.o AUTO HRVATSKA PRODAJNO SERVISNI CENTRI d.o.o AUTO HRVATSKA PRODAJNO SERVISNI CENTRI d.o.o AUTO-SAFIR d.o.o Auto Safir d.o.o BELINAMONT d.o.o CEE STROJEVI d.o.o ECCOS INŽENERING d.o.o ELEKTROCENTAR PETEK d.o.o ELTRA MG GOLUBOVEČKI KAMENOLOMI d.o.o Habeić doo KONČAR - INEM d.o.o MAN IMPORTER HRVATSKA d.o.o MAN IMPORTER HRVATSKA d.o.o MB SERVIS, Obrt za održavanje i popravak motornih vozila, strojeva, opreme i trgovinu, vlasnik Davor Žaler MIKRA MATIK AUTODIJELOVI d. o. o PASTOR TVA d.d R-PIM d.o.o ROBERT BERGER vl.obrta "BERGER"

67 Tahograf d.o.o TED d.o.o TRA-MONT d.o.o WELLMAX d.o.o X-PANEL d.o.o Zaštita i sigurnost d.o.o ADA-SERVIS d.o.o ĆIBO-PROMET d.o.o INTERPART SP d.o.o LAKMUS d.o.o SAVA-PROMET d.o.o SREBRA SYSTEM d.o.o Tehmar d.o.o ŽUPANIJSKE CESTE ZAGREBAČKE ŽUPANIJE d.o.o GIP PIONIR d.o.o Vodograd-I.G VODOTEHNIKA d.d GRADATIN d.o.o TROL DK d.o.o TROL DK d.o.o Industrooprema d.o.o Voith Turbo d.o.o ENERGOREMONT,d.d PELMEN d.o.o Agro-Honor d.o.o AUTO-MAG d.o.o AUTO-MAG d.o.o

68 GRA-PO d.o.o GRA-PO d.o.o SAMOBORKA D.D SAMOBORKA D.D Bilo Zagreb d.o.o CONTROLMATIK d.o.o DALEKOVOD-PROIZVODNJA d.o.o DRAŽEN KOVAČIĆ, vl. obrta "SERVIS IMP CRPKE" ELEKTROCENTAR PETEK d.o.o TERRA JASKA d.o.o Biromax d.o.o PEEK PROMET d.o.o PEEK PROMET d.o.o P.G.P. d.o.o DELTRON d.o.o DELTRON d.o.o PUGAR d.o.o FANOS d.o.o GRADATIN d.o.o MBM d.o.o MBM d.o.o Industrooprema d.o.o Industrooprema d.o.o VODOSKOK d.d VODOSKOK d.d VODOSKOK d.d M.B. AUTO d.o.o

69 TOI TOI d.o.o DILJEXPORT d.o.o KEFO d.o.o AUTO-MAG d.o.o AUTO-MAG d.o.o AUTO-MAG d.o.o GRA-PO d.o.o Kuna Corporation d.o.o MAXMAR GRUPA D.O.O RASCO d.o.o SMIT-COMMERCE d.o.o AnAs d.o.o AUTO HRVATSKA PRODAJNO SERVISNI CENTRI d.o.o BERGER ELEKTROMEHANIKA vl. Robert Berger CIAK TRUCK d.o.o Eccos-inženjering d.o.o MAN IMPORTER HRVATSKA d.o.o MIKRA MATIK AUTODIJELOVI d. o. o NIVAG EXPORT d.o.o REDOX Tegra d.o.o BIZMUT D.O.O KOVING D.O.O TEMEX d.o.o.; GIP PIONIR d.o.o.; GEAMEDITOR d.o.o INSTAL-PROM d.o.o TERMORAD d.o.o Peek promet d.o.o

70 Deltron d.o.o VIATOR d.o.o VIATOR d.o.o ORYX GRUPA d.o.o ORYX GRUPA d.o.o Mato el-d d.o.o Vodotehnika d.d Elektrokem d.o.o Elektrokem d.o.o Elektrokem d.o.o Fanos d.o.o Gradatin d.o.o Semafor d.o.o Ro tehnologija doo MBM d.o.o Industrooprema d.o.o Industrooprema d.o.o AUTOBUS d.o.o Autobus d.o.o AUTOBUS d.o.o Vodoskok d.d FDS-TRGOVINA d.o.o FDS-TRGOVINA d.o.o USLUGA d.o.o Voith Turbo d.o.o TOKOS d.o.o ENERGOREMONT,d.d

71 C.I.A.K. d.o.o C.I.A.K. d.o.o Kemokop d.o.o Kemokop d.o.o PELMEN d.o.o., Zagreb, Garićgradska Auto Benussi d.o.o AUTO ENIGMA d.o.o AUTO-MAG D.O.O GRADITELJ SVRATIŠTA d.o.o., Zagreb, Ivana Česmičkog HENNLICH industrijska tehnika KONČAR-ELEKTRONIKA I INFORMATIKA d.d KUDUMIJA TRADE OBRT SERVIS IMP CRPKE, vl. Darko Kovačić Smit-Commerce d.o.o ZAGREBAČKO ELEKTROTEHNIČKO PODUZEĆE D.D Zagrel Rittmeyer d.o.o ELTRA MG d.o.o KOVING D.O.O MENDIS-PROJEKT d.o.o MENDIS-PROJEKT d.o.o Metis d.d MILENIUM TRADE D.O.O Vodoplast promet d.o.o BOLČEVIĆ-GRADNJA d.o.o., Sesvetski Kraljevec, Dugoselska 57, MGV d.o.o., Zagreb, Slimska PEEK PROMET d.o.o Vodotehnika d.d INDUSTROOPREMA d.o.o

72 I.B. JAZBINA d.o.o VODOSKOK d.d ENERGOREMONT,d.d ENERGOREMONT,d.d Elektrocentar Petek d.o.o Elektrocentar Petek d.o.o KOR d.o.o TISAK DA-DA d.o.o TISAK DA-DA d.o.o VELEKEM d.d ELICOM d.o.o GTM d.o.o GEORAD d.o.o GIP PIONIR d.o.o HRT-ŠARIĆ d.o.o Hvar d.o.o SIGNALGRAD d.o.o TITAN CONSTRUCTA d.o.o AMB gradnja d.o.o VODOSKOK d.d VODOSKOK d.d GUMIIMPEX-GRP d.d GUT AUTO-MAG d.o.o AUTO-MAG d.o.o BOLČEVIĆ-GRADNJA d.o.o Graditelj svratišta d.o.o

73 KOM - TRADE d.o.o M.Soldo d.o.o POLJOOPSKRBA-TEHNO d.d POLJOOPSKRBA-TEHNO d.d Prigorac-građenje d.o.o SMIT - COMMERCE d.o.o SMIT-COMMERCE d.o.o Šušković-građenje d.o.o Trgometal d.o.o ZAGORJE GRADNJA d.o.o Anas d.o.o AUTO HRVATSKA d.d CONTROLMATIK d.o.o MAN IMPORTER HRVATSKA d.o.o MAN IMPORTER HRVATSKA d.o.o TI KEM d.o.o EKO - MIKS d.o.o Gutta Hrvatska d.o.o ITT - Rijeka d.o.o PALIĆ INŽENJERING d.o.o Turković d.o.o VEKTRA d.o.o Zajednica ponuditelja SOKOL MARIĆ d.o.o., BILIĆ-ERIĆ d.o.o. i V GRUPA d.o.o Zajednica ponuditelja SOKOL MARIĆ d.o.o., BILIĆ-ERIĆ d.o.o. i V GRUPA d.o.o Narodne novine d.d.; Tip-Zagreb d.o.o.; ZVIBOR d.o.o NOVI URED d.o.o.; STUBLIĆ IMPEX d.o.o MBM d.o.o

74 Industrooprema d.o.o Industrooprema d.o.o Industrooprema d.o.o VODOSKOK d.d VODOSKOK d.d EOL GRUPA d.o.o OMNIMERKUR d.o.o Trgometal d.o.o BENNINGHOVEN GmbH & Co. KG BENNINGHOVEN GmbH & Co. KG KOMOP d.o.o METALNO PLASTIČNA GALANTERIJA Kamenolom Gorjak d.o.o.; GOLUBOVEČKI KAMENOLOMI d.o.o.; HOLCIM MINERALNI AGREGATI d.o.o PEEK PROMET d.o.o GTM d.o.o GEORAD d.o.o GEORAD d.o.o TIGRA d.o.o Hvar d.o.o Pismorad d.d PAMAJO d.o.o VODOTEHNIKA d.d ELEKTROKEM d.o.o DETA PRUT d.o.o MBM d.o.o Industrooprema d.o.o VODOSKOK d.d

75 M.B. AUTO d.o.o GUT d.o.o O-K-TEH d.o.o ARBORI CULTURA d.o.o ELEKTRO-KOMUNIKACIJE d.o.o M. SOLDO d.o.o PA-EL d.o.o Prigorac-građenje d.o.o SMIT-COMMERCE d.o.o Šušković-građenje d.o.o URIHO - Ustanova za profesionalnu rehabilitaciju i zapošljavanje osoba s invaliditetom Elektrocentar petek d.o.o HIDRAULIKA KURELJA d.o.o HIDROMEHANIKA d.o.o Končar - Električna vozila d.d Zagreb plakat.d.o.o EKO-MIKS d.o.o HIDROCOM d.o.o NISKOGRADNJA DONJI JALŠEVAC d.o.o PALIĆ INŽENJERING d.o.o Turković d.o.o Vodopromet d.o.o KING ICT d.o.o.; Info-kod d.o.o.; MR servis d.o.o TERMORAD d.o.o GEORAD d.o.o Deltron d.o.o PUGAR d.o.o

76 PISMORAD D.O.O SIGNALGRAD d.o.o Deta prut d.o.o Gradatin d.o.o EOL grupa d.o.o TEHNIX d.o.o AUTO-MAG D.O.O BOLČEVIĆ-GRADNJA D.O.O ELEKTROCENTAR Petek, d.o.o Graditelj svratišta d.o.o AUTO HRVATSKA Prodajno Servisni Centri d.o.o CHROMOS FRAGARIA PLANTA d.o.o MALI GRM d.o.o Oktal Pharma d.o.o Source: CRCB Notes: sb[=1] : tender with single bidder x2 [=1] : the net contract value is more than twice of the estimated value x3 [=1] : the net contract value is more than 3 times higher than the estimated value x5 [=1] : the net contract value is more than 5 times higher than the estimated value 76

77 Appendix 2: Distribution of main variables Figure A2.1.: The histogram of logarithm of net contract value (HRK), , N = 5,922 Source: CRCB own calculation based on data of EPRCRC Figure A2.2.: The histogram of logarithm of estimated net contract value (HRK), , N = 4,653 Source: CRCB own calculation based on data of EPRCRC 77

78 Figure A2.3.: The histogram of logarithm of net contract value of tenders without competition (million HRK), , N = 1,443 Source: CRCB own calculation based on data of EPRCRC Figure A2.4.: The histogram of Competitive Intensity (ICI), , N = 4,238 Source: CRCB own calculation based on data of EPRCRC 78